REMOTE SENSING FOR SETTLEMENT PATTERNS IN METROPOLITAN REGIONS:
THE CASE OF NATIONAL CAPITAL REGION (NCR) OF DELHI, INDIA
Mahavir
Centre for Remote Sensing, School of Planning and Architecture, 4-B, I. P. Estate, New Delhi 110002, INDIA
mahavir57@yahoo.com
KEY WORDS: Settlement Patterns, Metropolitan Regions, Remote Sensing, Continuously Built-up Area, NCR, Delhi
ABSTRACT:
The author earlier conducted a research, attempting to make two models, one for understanding, monitoring and describing settlement
patterns in metropolitan regions, dependant primarily on remotely sensed data; and the second, for predicting settlement patterns in such
regions, using National Capital Region (NCR), of Delhi, India, as a case study. As part of the model, a transferable approach of the
‘Continuously Built-up Area’ (CBA) was adopted, which allows identification of settlements at different levels of generalization. With the
capability of providing frequent and synoptic views over large areas, remotely sensed data (e.g., satellite images) proved to be a useful base
for carrying out macro-analyses of settlement patterns, at frequent intervals. The model for prediction was applied for the target year of
2001. About 10 years have passed by since the base year of data collection. While the region has changed tremendously during these years,
recent data (Census of India, 2001) pertaining to the then target year (2001) is also now available. The moment is also apt, keeping in mind
that the Regional Plan – 2001 National Capital Region has now expired and the Draft Regional Plan - 2021 has just come in. This paper
reviews the models for 2001.
It has been found that the predictions made by the model with regard to settlements becoming `towns’, CBAs and couples have largely come
true. It also confirms the role satellite images can play in reasonably predicting settlement patterns in metropolitan regions.
1.
1.1
INTRODUCTION
Modelling Settlement Patterns
Analysis of settlement patterns is useful in making decisions
about the location of new investments, and about the
potential for clustering services and facilities in new ways to
increase the capacity of human settlements to stimulate
development in their areas. Settlement patterns in and around
metropolitan regions become all the more important when
considering the pace with which the urban settlements are
increasing, both in number and size. In addition, in
developing countries, there is a new category: increasingly
large numbers of spontaneous human settlements are coming
into existence on the urban fringes.
Left to ‘natural’ processes, these phenomena produce
inefficient as well as ineffective settlement patterns. To
reduce the social, economic and environmental cost of
inefficiency, management is introduced to address problems
arising from the processes. An attempt was made earlier, by
this author to suggest a model for describing the settlement
pattern, and for exploring alternatives for the management of
the future development of that pattern.
No sufficiently comprehensive methods exist for defining
and recognizing spatial patterns. The methods available (e.g.,
the spatial autocorrelation) are generally for point patterns,
which are best analyzed partly visually and partly manually.
Cluster analysis techniques are also not suitable for spatial
clusters.
A number of theories and models have been put forward to
describe and understand human settlement patterns. One
important conclusion that can be drawn from a review of
these models is that they all help in analyzing the existing
patterns. Most of them help, in a limited way, in predicting
settlement patterns and rarely, in planning settlement pattern
changes. Moreover, the prevalent models consider
settlements as point locations when analyzing their spatial
distribution. The size of the settlement, not in terms of the
population it holds but in terms of the area it covers on
ground, or its shape, is hardly ever considered.
1.2
Continuously Built-up Area
In considering urban growth, one of the first questions to be
resolved was how to identify the changing extent of the
settlements. In response, a transferable approach of the
‘Continuously Built-up Area’ (CBA) was devised, which
allowed identification of settlements at different levels of
generalization. On the images at a scale of 1:250,000, the
minimum curtilage for delineation of CBA was 4mm x 4mm
(terrain 1km x 1km or 100 ha) and a minimum width of 2mm
(terrain 500m) was adopted. An `exclusive’ definition was
applied: meaning that in case of doubt, it was decided to be
non built-up. The concept encompassed suburban
settlements, industrial complexes, airports, etc. In this
approach, the reality of the moving urban edge rather than
municipal, administrative or political boundaries is seen
(e.g., through satellite images), recorded and continuously
updated so as to provide real time information to support
decision-making. The strength of the tool (i.e., the concept of
CBA, applied to remotely sensed data) is that it may be
calibrated to any level of generalization, corresponding to
different scales of mapped information and decision-making.
Another conclusion drawn is that most quantitative models of
settlement patterns rely heavily on assumptions such as
isotropic planes and rational economic behaviour. The
geometrically and statistically guided models do not really
work in the light of ‘urban managerialism’, political climate
and planning behaviour. Thus, methods will have to be
developed which are flexible not only in the conceptual
context (e.g., CBA) but also in their application to model
settlement pattern in a region. Satisfactory techniques are not
available for performing a ‘formal’ spatial pattern analysis
exercise. The methods developed have to be based on a
partial visual/ manual analysis.
Methods based on traditional concepts require accurate,
sophisticated and timely availability of data, both spatial and
attribute, on variety of subjects. The level of detail required
for information on many functions changes frequently and is
difficult to update on a regular basis, more so in developing
countries. Models lose their sophistication when outdated
data are fed into them. Further, in many urban and regional
systems it is important to ascertain the overall situation. But
the information we usually have, or are likely to have,
relates, by and large, to individual decisions.
With the capability of providing synoptic views over large
areas, remotely sensed data (e.g., satellite images) can be a
useful base for carrying out macro-analyses of settlement
patterns, at frequent intervals. It not only provides
information more efficiently than field surveys; it also offers
a new perception and a new understanding of human
settlements.
2.
MODELS
Model(s) were developed for describing and making
predictions and recommendations for settlement patterns
based on the new understanding of settlements. Through a
flexible approach in defining human settlements, and the use
of images from satellites, the model(s) are easily applied in
‘data-poor’ situations and with not well-established practices
of settlement pattern planning, as is the case in many
developing countries. It literally provides a tool for better
defining and comprehending the term ‘local’, in the context
of the 74th Amendment to the Constitution of India, designed
to strengthen the role of urban local bodies in promoting
local and economic development, and in improving the
environment. The two models, one for understanding,
monitoring and describing settlement patterns in
metropolitan regions, dependant primarily on remotely
sensed data (Fig. 1); and the second, for predicting settlement
patterns in such regions (Fig. 2) were developed. Both the
models are suggested to be recalibrated each time the results
of a new census become available or when a new plan for the
region is being prepared. However, parts of the models that
are dependant on remotely sensed data are suggested to be
adjusted more frequently. The models were then applied to
the National Capital Region (NCR) of Delhi, India, as
described below.
3.
Pradesh. It is the biggest and generally accepted to the bestdefined metro region in India. The region accommodated a
population of more than 37 million people (in 2001), of
which about 14 million lived in Delhi alone. Density of
population of the region is about 1,200 persons per sq. km,
with Delhi being the highest at more than 9,000 persons per
sq. km. About 56 per cent population of the Region is urban,
accommodated in 110 towns, including three metropolitan
cities, viz., Delhi, Meerut and Faridabad. The city of
Ghaziabad together with Loni is also emerging as a
metropolitan complex. Besides, there are about 7,950 rural
settlements, accommodating a rural population of 16.4
million. Plan policies for the Region (2001) were covered
under three Policy Zones, viz. NCT Delhi, Delhi
Metropolitan Area (DMA) and the Rest of the NCR.
However, the Draft Regional Plan – 2021 redesignated the
DMA as Central NCR and added Highway Corridor as
another policy zone.
3.1
Application of the Models on the NCR
The models as described in 2 above were applied to the NCR
(as in 1991). Extensive use of satellite images from IRS-1A
and 1-B (LISS-I and LISS-II) at resolutions 72.5m and
36.25m was made for the purpose. The images covered two
time periods, e.g. 1990-91 (corresponding to the Census
year) and 1993-94 (corresponding to the ground truth year).
All products were standard CCT, 6250BPI, BSQ format. The
image interpretation was primarily visual, as indicated in the
models above. The models, drawing heavily from the
operational definition of `CBA’, examined the images for
identifying existing `assemblage’ of human settlements,
`couples’, `insular’ settlements likely to join an assemblage
or to form a new assemblage/ couple, existing villages with a
potential to become urban by 2001 and existing and potential
towns that were likely to stay as `insular’. While predicting a
scenario for the year 2001 (target year for the then Regional
Plan), 4 `assemblage’ settlements, 11 `couple’ settlements, 3
`functional couple’ settlements and 65 `insular’ settlements
were identified. Population for these settlements was also
projected for the year 2001, primarily using the `CBA’ as
interpreted on the satellite images and supported with their
present (then) population, etc.
The population figures now announced by the Census for the
year 2001 and the analysis of the settlement pattern as
presented in the Draft Regional Plan for the year 2021
provided encouraging results and confirm many of the
predictions made by the models. Satellite images from IRS1C/1D (LISS III + PAN) of the year 1999, as interpreted by
the NCR Planning Board, were also used for making these
observations. The results are described in following
paragraphs.
NCR OF DELHI
4.
National Capital Region (NCR) of Delhi, India is a complex
entity, covering an area of about 33,578 sq. km. lying
between 27003’N and 29029’N latitude and 76007’E and
78029’E longitude. It is a complex planning and
administrative system. More complex is the planning for
settlement pattern for the region, with human settlements in
population sizes ranging from less than 500 persons to over
14 million people. The Region surrounds and includes the
city/ state of Delhi (formally known as National Capital
Territory of Delhi or the NCT Delhi). It also covers parts of
the states (provinces) of Haryana, Rajasthan and Uttar
4.1
APPRAISAL OF THE MODELS
Potential Towns
As per the definition adopted by Census of India, towns are
places with a municipal corporation, a municipal committee,
a town committee, a notified area committee or a cantonment
board; also all places having a population of more than 5,000
persons, a density of population of 400 persons per sq. km.,
pronounced urban characteristics and at least three-fourth of
its adult male population employed in pursuits other than
agriculture. A town can be declassified, however, if one of
criteria is not met subsequently. As described in 3.1 above,
the villages (in 1991) having a potential to become towns (in
2001) were identified. The decision was based on their (then)
population, growth rate, population density, percentage of
male working population in non-agricultural activities,
besides `CBA’. This list included the (then) villages of
Dharoti Khurd, Salarpur Khadar, Nahri, Ladrawan, Ramgarh,
Dunda Heda, Rewari (rural), Kishangarh, Asan Khurd,
Pinagwan, Sidhrawali, Khandsa, Tapookra, Sahoti,
Mohiuddinpur, Tilpat, Aminagar (excluding the villages
within the NCT of Delhi, which were all expected to be
urbanised by 2001). In addition, a few combination of
villages were also expected to become urban, purely on the
basis of `CBA’, although they failed miserably in atleast one
criteria essential for declaring a settlement `town’ by the
Census. These were of Atseni, Bait and Bihuni; Uchagaon
and Amargarh; Bahin and Kot; Bahalgarh and Rathdhana;
and Shahajahanpur.
It is very encouraging to note that 15 of the above 22 rural
settlements indeed became `towns’ in the year 2001. Out of
the remaining 7, four did join other `urban agglomerations’
and therefore did not appear as separate towns. The
remaining 3 did not become towns as they were predicted
purely on the basis of `CBA’, as interpreted on the image and
no consideration to other data was given.
4.2
`Assemblage’ Settlements
The models also predicted four `assemblage’ settlements to
be there in 2001. The first `assemblage’, primarily of NCT
Delhi included the settlements of Loni, Western Ghaziabad,
Noida, Tilpat, Faridabad, Gurgaon, Bahadurgarh, etc. The
combined population for the assemblage was projected at
15.5 million by the year 2001. Another `assemblage’ was
that of towns of Eastern Ghaziabad, Dasna and Dunda Heda,
Operational
Definition of
Continuously
Built-up Area
Existing Census
Towns
Locate on TopoSheets; Obtain
X, Y
Coordinates
Use X, Y
Coordinates for
orientation of
Images
Digital Raw
Data of the
Satellite Images
of the Region
Image
Processing and
Enhancement on
zoomed in
scenes of the
Existing/
Potential
Towns/
Settlements, etc.
FCC of SubScenes/ Major
Settlements
Delineation of Continuously Built-up
Area for each settlement or assemblage,
etc.
Apply various Point Pattern Recognition
and Analysis Techniques
Apply various Area Pattern Recognition
and Analysis Techniques
Census/ Other
Thematic data
Developing Better Understanding of the
Existing Settlement Pattern/Change in
Pattern
Socio-Economic/
Political
Considerations
Process(s) recommended
to be repeated
periodically
Rules on Min.
Curtilage,
Minimum
Width,
Idealisation,
Generalisation
Assemblage, etc.
Input for Model for Predicting and
Recommending
The overall process recommended to be
repeated whenever results of a new census
made available, or a new plan is under
preparation.
Fig. 1
Model for Understanding, Monitoring and Describing Settlement Patterns
Digital raw data of
the
satellite
images of the
region
Definition of
Towns
Criteria of
Population,
Density,
Workforce,
Economic
Importance,
Function etc.
Existing Census
Towns
Locate on Topo
Sheets; Obtain
X, Y
Coordinates
Use X, Y
Coordinates for
orientation of
Images
Digital
Raw
Data of the
Satellite Images
of the Region
Image
Processing and
Enhancement
on zoomed in
scenes of the
Existing/
Potential
Towns/
Settlements
FCC of SubScenes/ Major
Settlements
Apply various Point Pattern Recognition
and Analysis Techniques
Census/ Other
Thematic data
SocioEconomic/
Political
Considerations
Process(s)
recommended to be
repeated periodically
Rules of
Assemblage,
Couple, etc.
Existing Census Villages
Potential Towns
for future period
Potential
`Settlements’ for
future period
Operational
Definition of
Continuously
Built up Area
Locate on Topo-sheets, Identify the
Settlements; Obtain X,Y Coordinates;
Obtain relevant attribute data for existing
towns
Delineation of Continuously Built-up
Area for each settlement or assemblage,
etc.
Rules on Min.
Curtilage,
Min. Width,
Idealisation,
Generalisation
Assemblage,
etc.
Apply various Area Pattern Recognition
and Analysis Techniques
Take policy decisions regarding
Continuation/ Modification/ Alteration of
the Settlement Pattern
Generate a pattern of Continuously Builtup area/ Assemblage/ Couples, etc.
The overall process recommended to be
repeated whenever results of a new
census made available, or a new plan is
under preparation.
Fig. 2
FCC of SubScenes/
Settlements
Re-aggregate relevant attribute data;
Assign settlement hierarchy, population,
functions, services, etc.
Model for Predicting and Recommending Settlement Patterns
Input from Model
for
Understanding,
Monitoring and
Describing
Rules for
potential
CBAs/
Assemblage/
Couples
with a combined projected population of 0.4 million. It may
be noted that, in reality, there is no entity as Western or
Eastern Ghaziabad. The distinction was made by this author
only for the fact that river Hindon makes the division with
only one bridge to join the two parts. If the division by the
river is ignored, these `assemblages’ will make one large
assemblage with a combined projected population of 15.9
million persons in 2001.
It is heartening to know that the settlements for the above
assemblage now total up to 16.5 million. Moreover, the
settlements are primarily being treated as Central NCR
(erstwhile DMA). These settlements mainly compose of
Ghaziabad, Modinagar, Noida, Faridabad, Gurgaon,
Bahadurgarh, Sonepat and Kundli. It may also be noted that
the division between these settlements among themselves
and with NCT Delhi is, in reality, an administrative and
political one and they in fact constitute one large `CBA’.
4.3
`Couple’ Settlements
As stated earlier, the models also identified a few `couple’
settlements. These included the couples of Alwar-Ramgarh,
Hapur-Babugarh,
Surajpur-Kasna,
Sonipat-Bahalgarh,
Baghpat-Agarwal Mandi, Kithor-Shahajanpur, KhairthalKishangarh, Hailey Mandi-Pataudi, Tikri-Doghat, DauralLawar and Bahsuma-Hastinapur. Functional couples of
Modinagar–Mohiuddinpur/ Rori – Meerut, SikandrabadBulandshahar–Khurja; and Rewari–Dharu Hera/ Bhiwadi
were also identified.
Images for the year 1999 (nearly corresponding to the year of
Census, 2001) confirm the above couples taking place. It is
very satisfactory to note that the Draft Regional Plan – 2021
indeed proposes to develop many of these as `integrated
complexes’. They include Baghpat-Baraut, Ghaziabad-Loni,
Faridabad-Ballabgarh,
Bulandshahar-Khurja,
BehrorShahjahanpur-Neemrana, Rewari-Dharuhera-Bawal and
Sonepat-Kundli. Proposed Land Use plan for the year 2021
also indicates the willingness of the NCR Planning Board to
accept these assemblages, couples, etc., except for the fact
that these are still separate entities for the purpose of
administration as well as Census.
decision makers to gain better understanding of this
complexity, in a way that can be impartially, and with
consistency, executed by technicians.
An interesting finding was that working at the metropolitan
regional scale, images with low spatial resolution are actually
more useful than images with higher resolution. The
remotely sensed data with spatial resolution of 72.5m is
found appropriate for the purpose of this task (e.g., modelling
settlement patterns), providing a synoptic view of the entire
region under study. Specifically, its comparatively low
resolution performs the otherwise difficult task of
determining the ‘built-up’ pixel out of a mix of 64 pixels (of
10m resolution compared to 80m) in reality.
It may be noted that although, in principle, the method can be
extended to rural settlements too, it will require higher
resolution data and detailed ground truth. Similarly, the need
for support from other traditional data sources cannot be
totally ruled out as the images fail to deduce functions and
interaction between human settlements.
REFERENCES
Bolt, D., 1996. An urban community strategy for
implementing decentralisation and devolution policies. ITC
Journal, 1996-1, pp.18-28.
Census of India, 2001.
Mahavir, 1996. Modelling Settlement Patterns for
Metropolitan Regions: Inputs from Remote Sensing. ITC,
Enschede, The Netherlands.
NCRPB, 1988. Regional Plan – 2001. National Capital
Region Planning Board, Ministry of Urban Development,
Government of India, New Delhi.
NCRPB, 2004. Draft Regional Plan – 2021. National Capital
Region Planning Board, Ministry of Urban Development,
Government of India, New Delhi.
ACKNOWLEDGEMENTS
5.
CONCLUSIONS
From the above review, it may be concluded that the models,
depending heavily on the remotely sensed data, as applied to
the concept of `CBA’ provide an excellent framework for
understanding, monitoring and describing existing patterns,
and for making predictions and recommendations. Moreover
the methods provide for periodic revisions and updates
corresponding to availability of two major sources of data
(i.e., the census and the satellite images). An important
learning from the review is that the identification of rural
settlements having potential to become urban may not be
guided totally by the `CBA’ as seen on the images. At the
same time, this strengthens the concept of CBA rather than
discriminating the settlements as `towns’ or `villages’. The
use of satellite images, as illustrated in the case of NCR, also
confirms that the settlements will continue to grow
irrespective of their `administrative’ jurisdictions.
The concept of CBA, supported with tools such as remote
sensing and GISs, enables planners, administrators and
S. Surendra, Swati Khanna, Inshu Minocha, Kiran Punia.